MIMO Detection Methods: How They Work [Lecture Notes]
Summary (2 min read)
RELEVANCE
- The most important motivating application for the discussion here is receivers for multiple-antenna systems such as multiple-input, multiple-output (MIMO), where several transmit antennas simultaneously send different data streams.
- Essentially the same problem occurs in systems where the channel itself introduces time- or frequency-dispersion, in multiuser detection, and in cancellation of crosstalk.
PROBLEM STATEMENT
- For simplicity of their discussion, the authors assume that all quantities are real-valued.
- If H has structure, for example, if it is a Toeplitz matrix, then one should use algorithms that can exploit this structure.
- (2) Problem (2) is a finite-alphabet-constrained least-squares (LS) problem, which is known to be nondeterministic polynomial-time (NP)-hard.
SOLUTIONS
- (4) Problem (4) can be visualized as a decision tree with n1 1 layers, |S| branches emanating from each nonleaf node, and |S|n leaf nodes.
- Finally, to each node, the authors associate the symbols 5s1, c, sk6 it takes to reach there from the root.
- Clearly, a naive but valid way of solving (4) would be to traverse the entire tree to find the leaf node with the smallest cumulative metric.
- Such a brute-force search is extremely inefficient, since there are |S|n leaf nodes to examine.
- The authors will now review some efficient, popular, but approximate solutions to (4).
ZF DETECTOR WITH DECISION FEEDBACK (ZF-DF)
- Consider again ZF, and suppose the authors use Gaussian elimination to compute s| in (5).
- In the decision-tree perspective, ZF-DF can be understood as just examining one single path down from the root.
- Clearly, after n steps the authors end up at one of the leaf nodes, but not necessarily in the one with the smallest cumulative metric.
- In its simplest form (as explained above), ZF-DF detects sk in the natural order, but this is not optimal.
- Even with the optimal ordering, error propagation severely limits the performance.
SPHERE DECODING (SD)
- The SD [2], [9] first selects a user parameter R, called the sphere radius.
- Effectively, the authors will adaptively prune the decision tree, and visit much fewer nodes than those in the original sphere.
- In particular, the algorithm does not examine any branches stemming from the node “5” in the right subtree.
- The SD algorithm can be improved in many other ways, too.
- The symbols can be sorted in an arbitrary order, and this order can be optimized.
FIXED-COMPLEXITY SPHERE DECODER (FCSD)
- FCSD [3] is, strictly speaking, not really sphere decoding, but rather a clever combination of brute-force enumeration and a low-complexity, approximate detector.
- Authorized licensed use limited to: Linkoping Universitetsbibliotek.
- To form its symbol decisions, FCSD selects the leaf, among the leaves it has visited, which has the smallest cumulative metric f1 1s1 2 1c1 fn 1s1, c, sn 2 .
- FCSD solves (2) with high probability even for small r, it runs in constant time, and it has a natural parallel structure.
SEMIDEFINITE-RELAXATION (SDR) DETECTOR
- The idea behind SDR [5], [6] is to relax the finite-alphabet constraint on s into a matrix inequality and then use semidefinite programming to solve the resulting problem.
- The authors explain how it works, for binary phase-shift keying (BPSK) symbols (sk [ 5616).
- It then proceeds by minimizing Trace {CS} with respect to S, but relaxes the rank constraint and instead requires that S be positive semidefinite.
- This relaxed problem is convex, and can be efficiently solved using so-called interior point methods.
- The error incurred by the relaxation is generally small.
LATTICE REDUCTION (LR) AIDED DETECTION
- The idea behind LR [8], [9] is to transform the problem into a domain where the effective channel matrix is better conditioned than the original one.
- Naturally, there are many such matrices (T5 6 I is one trivial example).
- Namely, some of its elements may be beyond the borders of the original constellation.
- Hence a clipping-type operation is necessary and this will introduce some loss.
SOFT DECISIONS
- It is then of interest to take decisions on the individual bits bk,i, and often, also to quantify how reliable these decisions are.
- Fortunately, (9) can be relatively well approximated by replacing the two sums in (9) with their largest terms.
- This is naturally accomplished by many of the methods the authors discussed, by simply including the terms corresponding to all leaf nodes in the decision tree that the algorithm has visited.
CONCLUSIONS
- This depends much on the purpose of solving (2): what error rate can be tolerated, what is the ultimate measure of performance (e.g., frame-error-rate, worst-case complexity, or average complexity), and what computational platform is used.
- This complicates the picture, because notions that are important in slow fading (such as spatial diversity) are less important in fast fading, where diversity is provided anyway by time variations.
- Detection for MIMO has been an active field for more than ten years, and this research will probably continue for some time.
Did you find this useful? Give us your feedback
Citations
5,158 citations
2,717 citations
626 citations
526 citations
244 citations
References
2,291 citations
1,791 citations
"MIMO Detection Methods: How They Wo..." refers background in this paper
...For each node on layer r, the algorithm considers 5s 1 , ..., s r 6 fixed and formulates and solves the subproblem min ) 1 5 2 1 2 3 4 1 3 4 3 1 1 9 1 5 3 2 7 8 7 4 5 6 1 0 8 9 17 ZF-DF 1 5 2 1 2 3 4 1 3 4 3 1 1 9 1 5 3 2 7 8 7 4 5 6 1 0 8 9 17 SD, No Pruning (here: R = 6) 1 5 2 1 2 3 4 1 3 4 3 1 1 9 1 5 3 2 7 8 7 4 5 6 10 8 9 17 SD, Pruning (here: R = ∞) 1 5 2 1 2 3 4 1 3 4 3 1 1 9 1 5 3 2 7 8 7 4 5 6 10 8 9...
[...]
...the ultimate decisions on s. (a) ZF-DF: At each node, the symbol decision is based on choosing the branch with the smallest branch metric....
[...]
...(8) Problem (8) is comparatively easy, since HT is well conditioned, and simple methods like ZF or ZF-DF generally work well....
[...]
...FCSD does this approximately , using a low-complexity method (ZF or ZF-DF are good choices)....
[...]
...In its simplest form (as explained above), ZF-DF detects s k in the natural order, but this is not optimal....
[...]
1,760 citations
Additional excerpts
...The SD [2], [9] first selects a user parameter R, called the sphere radius....
[...]
1,616 citations
Additional excerpts
...The SD [2], [9] first selects a user parameter R, called the sphere radius....
[...]
966 citations
Related Papers (5)
Frequently Asked Questions (9)
Q2. What is the simplest way to extend the constellation S?
If the constellation S is uniform, then S may be extended to a scaled enumeration of all integers, and S n may be extended to a lattice S n.
Q3. What is the important motivating application for the discussion here?
The most important motivating application for the discussion here is receivers for multiple-antenna systems such as multiple-input, multiple-output (MIMO), where several transmit antennas simultaneously send different data streams.
Q4. What is the method for detecting a mIMO?
This depends much on the purpose of solving (2): what error rate can be tolerated, what is the ultimate measure of performance (e.g., frame-error-rate, worst-case complexity, or average complexity), and what computational platform is used.
Q5. What is the simplest way to find a matrix?
Once ŝ r is found, it is transformed back to the original coordinate system by taking ŝ5 T21ŝ r.LR contains two critical steps.
Q6. What is the problem LR solves first?
It then computesŝr ! arg min s9PSn || y2 1HT 2s r||. (8) Problem (8) is comparatively easy, since HT is well conditioned, and simple methods like ZF or ZF-DF generallywork well.
Q7. What is the constraint for the ZF detector?
The ZF detector first solves (2), neglecting the constraint s [ S ns| ! arg min s[Rn 7y2Hs 7 5 arg mins[Rn 7 y, 2 Ls 7 5 L21y,. (5)Of course, L21 does not need to be explicitly computed.
Q8. what is the probability that the transmitter sent s?
Such reliability information about a bit is called a “soft decision,” and is typically expressed via the probability ratioP 1bk,i5 1| y 2 P 1bk,i5 0| y 2 5 g s:bk,i 1s251 P 1s| y 2 g s:bk,i 1s250 P 1s| y 25 g s:bk,i 1s251expa2 1s || y2Hs||2bP 1s 2 g s:bk,i 1s250expa2 1s || y2Hs||2bP 1s 2 .(9)Here “s:bk,i 1s 2 5 b” means all s for which the ith bit of sk is equal to b, and P 1s 2 is the probability that the transmitter sent s.
Q9. What is the cost of finding T?
This is computationally expensive, but if the channel H stays constant for a long time then the cost of finding T may be shared between many instances of (2) and complexity is less of an issue.